Project checklist Flashcards
1.
- Frame the problem and look at the big picture
2.
- Get the data
3.
- Explore the data to gain insights
4.
- Prepare the data
5.
- Explore many different models and shortlist the best ones
6.
- Fine-tune your models and combine them into a great solution
7.
- Present your solution
8.
- Launch, monitor, and maintain your system
1.1
1.1 Understand business problem and objective, explore hidden assumptions
1.2
1.2 Current solution
1.3
1.3 Measure of success
2.1
2.1 Get access to data and create workspace
2.2
2.2 Consider legal and privacy obligations
2.3
2.3 Set test data aside
3.1
3.1 Check feature characteristics
3.2
3.2 Visualize the data
3.3
3.3 Study feature correlations
4.1
4.1 Data cleaning
4.2
4.2 Feature selection and engineering
4.3
4.3 Feature scaling
5.1
5.1 Train and compare many quick and dirty models
5.2
5.2 Quick feature selection and engineering, iterate
5.3
5.3 Shortlist 3-5 most promising models, preferring different errors
6.1
6.1 Fine tune hyper-parameters, optimize against business objective
6.2
6.2 Try ensemble method
6.3
6.3 Measure performance against test set
7.1
7.1 Document
7.2
7.2 Explain achievement of business objective
7.3
7.3 Present other interesting points
8.1
8.1 Make production ready
8.2
8.2 Write monitoring code
8.3
8.3 Set up retraining of model
